Multibay UPMP instances

This dataset provides instances for three experiments of the multibay unit-load pre-marshalling problem. For each experiment, we provide the instances as JSON-files and the warehouse layouts in a separate folder as CSV-files.

• Folder "Multibay_UPMP_instances_single_tier": Contains 3120 different multibay instances for a single tier. Each instance is repeated four times for different tie breaking variants "compar mode" (12480 files overall).
• Folder "Multibay_UPMP_instances_single_opposite_four_directions": Contains 240 different multibay instances for the comparison of different access directions. The warehouse layouts are provided in a separate folder. The layout files indicated with "opposite" use "-6" for tiles that can be used for travelling, but not to access a storage bay.
• Folder "Multibay_UPMP_instances_two_tiers": Contains 280 different multibay instances for two tiers.

The JSON-file for an instance contains:

• layout_file: Name of the used warehouse layout
• fill_level: Precise fill level
• max_priority: Number of priority groups
• height: Number of tiers
• comparison_order: Used tie breaking variant
• bay_info: Information for each bay includes x/y coordinates for the upper left corner of the bay, the width and length, the available access directions as well as access point IDs
• initial_state: 3-dimensional array for each bay with the priority groups per storage location. Zero is empty space.
•  access_points: Information for each access point with ID, direction, and stack x/y coordinates it is pointing to.

The JSON file names include the bay size after "Size", the warehouse layout after "Layout", the desired fill level (rounded) after "fill level", the number of priority groups after "max_p", and the tie breaking variant after "compar_mode".

The CSV-file for each layout contains:

• A series of numbers for each row that defines a grid
• The numbers are: -5 for travel space, 1 for storage space, and -6 for travel space that cannot be used to access a bay.